Everybody quotes LTV. Almost nobody calculates it correctly, and fewer still act on it. Most of what gets called an LTV number is a vanity figure: all-time revenue, every cohort blended together, no margin applied, no time window attached. It looks great on a slide and it cannot fund a single decision.
This is the operator’s version. The real formulas, an honest picture of the tooling, and the exact execution inside Klaviyo where the predictive data already sits. No theory padding.
The whole discipline fits in three moves. Measure it properly using cohorts, contribution margin and real time windows. Diagnose which lever is broken, whether that is AOV, frequency, retention or margin. Act on it inside Klaviyo, where the predictive segments and triggers already live. Everything below is those three moves in order.
To keep it concrete, the maths runs on one made-up brand throughout: Marlow Botanics, a UK supplements DTC on Klaviyo and Shopify. £40 AOV, 40% contribution margin (so £16 of margin per order), £22 to acquire a new customer, roughly a 3-month reorder cycle giving 4 orders in year one. Every figure here is a placeholder to calibrate against your own numbers, not a benchmark.
This assumes you run Klaviyo on a brand doing £1M+ and you know what a flow is and how attribution works.
1. What LTV actually is, and what founders get wrong
LTV is the value a customer generates across their relationship with your brand. Simple definition, endlessly abused metric, and the abuse starts with the word “value.”
Most founders mean revenue. Revenue LTV is flattering and useless for decisions. The number that funds decisions is contribution-margin LTV: revenue minus COGS, discounts, returns, shipping, fulfilment and payment fees. A £120 revenue LTV at 30% margin is £36 of actual money, and you cannot spend revenue on ads.
The second abuse is the missing time window. “Our LTV is £180” means nothing without a period and a cohort attached. brands quote “12-month cohort LTV” or “60-day LTV,” never a bare “lifetime” number, because a lifetime figure on a four-year-old brand blends in customers acquired in a different market at a different CAC.
Two different tools sit underneath this. Historic LTV is what customers have already spent: accurate, backward-looking, good for reporting and VIP tiering. Predictive LTV is a model’s estimate of future spend. In Klaviyo, Predicted CLV forecasts the next 365 days and Total CLV combines both. Historic tells you who your best customers were. Predicted tells you who they will be, which is the one you target and budget on.
The five mistakes to stop making: quoting revenue instead of margin (inflates the number two to three times and hides broken unit economics); attaching no time window; blending all cohorts into one average that lets strong old cohorts mask weakening new ones; applying the SaaS 3:1 rule blind when DTC repeat curves behave nothing like multi-year subscriptions; and treating LTV as fixed. LTV is the output of your retention programme. It moves when you move it. That is the entire point of the guide.
Marlow: the founder quotes “our LTV is £180,” all-time revenue, every cohort blended. The operator’s number is 12-month cohort margin LTV: 4 orders × £16 = £64. Roughly a third of the vanity figure, and the only version that can be set against a £22 CAC.
2. How to calculate it: three methods
Three methods, ascending usefulness. Most brands only ever need the first two done properly.
Method 1, the simple formula. AOV × orders per year × years retained × contribution margin %. Good for a first read and for board conversations, useless for spotting a trend. On Marlow: £40 × 4 × 1.5 years × 40% = £96 margin LTV, which against a £22 CAC leaves £74 of margin per customer to fund overhead and profit.
Method 2, cohort-based, the one that matters. Group customers by first-purchase month, then track cumulative margin per customer at 30, 60, 90, 180 and 365 days since first order. This is the only method that shows whether retention is actually working, because it compares like with like: June 2025 at day 90 against June 2024 at day 90. A healthy curve is still climbing at month 12. A curve that goes flat after month 3 is a one-and-done brand. You do not need to build anything: Shopify’s cohort report, Klaviyo’s CLV reporting, or a pivot table on your raw order export all do it. The tool matters far less than doing it monthly on margin-adjusted
numbers.
Method 3, predictive. Probabilistic models that estimate future purchases per customer. Do not build this yourself. Klaviyo already runs one, retrained at least weekly, on your own order data, and section 6 covers how to use it.
Four calculation traps to avoid: ignoring discounts, returns and shipping (use net revenue after returns, then apply margin, or high-return apparel brands get destroyed by gross-revenue LTV); survivorship framing (divide by the whole cohort, not just returners); uncapped lifetime windows (cap at 12 months, 24 for considered purchases, because money arriving in year three does not pay this quarter’s ad bill); and using blended CAC in the comparison (returning customers cost near zero to reactivate, so compare cohort LTV against new-customer CAC).
Marlow: the January 2026 cohort sits at £28 margin per customer by month 3. The January 2025 cohort was at £34 at the same age. Total revenue was up 40% year on year, so nobody noticed the retention curve weakening for five months. That is the whole case for cohort measurement.
3. LTV:CAC and payback: the numbers that fund growth
Two numbers decide whether you can afford to grow.
LTV:CAC asks whether the customer is worth the ad spend. Take 12-month margin LTV, divide by CAC. Marlow: 4 orders × £16 = £64 margin LTV, ÷ £22 CAC = 2.9:1. Healthy DTC sits between 2.5:1 and 4:1 on 12-month margin LTV against new-customer CAC. Below 2.5:1 each new customer barely covers their own acquisition. Persistently above 5:1 usually means you are under-spending on acquisition and handing growth to competitors. Ignore the SaaS 3:1 rule; it is built on multi-year subscription revenue.
Payback asks when the cash comes back, and it matters more than the ratio, because you buy inventory and ads with cash, not with 12-month projections. Payback is the months until accumulated margin covers CAC. Marlow’s £22 CAC is not covered by the first order’s £16 of margin; the second order clears it, and with a 3-month reorder cycle that lands around month 4. Consumables typically pay back in 1 to 6 months, considered purchases in 3 to 12+, subscription faster because orders arrive on fixed cycles. A 3:1 ratio with a 4-month payback and a 3:1 with a 14-month payback are not the same business. The first recycles cash into the next cohort three times faster.
The 60-day check is where brands blow themselves up. Before scaling spend, run the same maths at day 60. Marlow at day 60 has most customers on one order, some on two: roughly £20 of margin against £22 CAC, about 1:1, survivable because the money returns within two months. The classic failure is a brand seeing 2.9:1 on the 12-month view, tripling ad spend, and running out of cash in 90 days because at day 60 it was sitting at 0.4:1 with most of that LTV not yet arrived. The 12-month number tells you the model works. The 60-day number tells you how fast you can afford to run it. Make spend decisions on the shorter one.
Marlow: 2.9:1 at 12 months, payback at month 4, about 1:1 at day 60. Green light to scale, in steps sized to the 60-day return, not the 12-month projection.
4. The levers that actually move LTV
LTV has four inputs: AOV, purchase frequency, retention length, margin. Every retention activity pulls one of them, so the skill is picking the right one first.
The second purchase is the inflection point. The biggest single LTV event in most DTC brands is the move from one order to two, because orders 2, 3 and 4 carry near-zero acquisition cost, so every point of repeat rate converts to margin at a rate paid ads never match. Practitioner data consistently shows a 5-point lift in repeat rate out-earning a 10% AOV lift over 12 months. Category repeat rates within 24 months run from roughly 15-20% for home goods to around 38% for supplements, and that spread in repeat behaviour, not CAC, is why two brands with identical acquisition costs build completely different P&Ls.
Which lever to pull depends on the brand. Consumable and replenishable brands (supplements, beauty, pet, food) pull frequency and retention first, because the product creates natural reorder cycles and the job is making sure the reorder happens with you, on time; replenishment flows and subscription are the whole game. Considered-purchase brands (home, furniture, electronics) pull AOV and margin, because repeat is structurally low and slow, so bundles, warranties, accessories and cross-category expansion beat chasing frequency that is not there. Apparel and mixed pull second-purchase rate, attacking time-to-second-order. Heavy discounters in any category pull margin first, because if repeat only happens with a code you are renting retention, not building it.
Four leading indicators to watch weekly: 60-day repeat rate by cohort (the earliest reliable signal a cohort’s curve will climb or flatten); median time to second order (stretching month over month means your post-purchase programme is weakening); percentage of a cohort at 2+ orders by day 90 (the compounding base, worth more than any campaign revenue screenshot); and discount dependency of repeat orders (a rising share of second orders using a code means margin LTV is falling even while revenue LTV looks fine).
Marlow is consumable, so the lever is frequency and retention and the metric to attack is time to second order, currently a 92-day median against a 3-month product cycle, target under 75. The maths: five points of repeat rate is one extra second order per 20 customers, £16 of margin each at zero added CAC. No AOV project competes with that.
5. Tooling: what you need, what you don’t
At £1M to £10M you need one source of truth for cohort LTV and one activation layer, and you probably already own both.
For measurement, Shopify’s cohort analysis plus a spreadsheet (margin-adjusted in the sheet) is enough for most brands under about £5M. Start there. For activation, Klaviyo’s predictive analytics is non-negotiable and already in your stack; it is where LTV data becomes revenue, and section 6 covers it. Dedicated LTV tools like Lifetimely or Peel are worth buying when the spreadsheet ritual keeps getting skipped or you need LTV broken out by acquisition channel; they buy discipline, not new maths. Profit and attribution platforms in the Triple Whale or Northbeam class are justified only at meaningful multi-channel paid spend and are overkill below that.
The rule that saves the most money: do not buy a tool to avoid defining your contribution margin. Every platform asks you for COGS, shipping and fee inputs, and if those are wrong the dashboard is an expensive way to be confidently wrong. Define margin once, in writing, then feed it to whatever tool you use.
Marlow: Shopify cohort report plus a spreadsheet with the 40% margin definition written at the top, refreshed monthly, and Klaviyo predictive analytics as the activation layer. No dedicated LTV tool until a real decision gets blocked by not having channel-level LTV.
6. Tracking LTV in Klaviyo
Klaviyo builds a CLV model on your account data and retrains it at least weekly. Most accounts never touch the output.
To unlock it, the account needs 500+ customers who have placed an order, 180+ days of order history with orders in the last 30 days, and some customers with 3+ orders. One catch: if you send Placed Order events via custom API, the order value must pass through the $value field with the real amount or CLV calculates wrong. Standard Shopify integrations handle this automatically.
The properties you get are Historic CLV (past spend, for reporting and historic VIP tiers), Predicted CLV (forecast spend over the next 365 days, for forward-looking VIP tiers and budget allocation), Total CLV (both combined, for overall ranking), churn risk (probability from 0 to 1 of not purchasing again, for intervention flows and exclusions), Expected Date of Next Order or EDNO (predicted next purchase date, for replenishment and winback timing), and average time between orders (for calibrating flow delays).
One rule governs all of it: predictions are segment-level tools. A single profile showing 1.43 predicted orders means “one or two, probably.” Averaged across 5,000 profiles, the same maths becomes reliable enough to plan budgets on.
The segments to build this week, all using the “Predictive analytics about someone” condition: Predicted VIPs (Predicted CLV in your top band, thresholds pulled from a CSV export first); at risk, worth saving (churn risk above account average AND historic CLV above median); replenishment window (EDNO within 14 days); overdue (EDNO in the past, no order since); and low predicted spenders (below threshold, your candidates for bestseller and entry-offer campaigns rather than premium pushes).
Those segments map to a clean tiering. Top 5% by Predicted CLV are VIPs: early access, launches first, no discounts, surprise-and-delight, review asks. The next 15% are future VIPs: cross-sell to premium lines, loyalty enrolment, bundle upsells. The middle 60% are core: bestsellers, replenishment, second-purchase pushes. The bottom 20% (low predicted or high churn risk) get entry offers, reduced send frequency, and suppression from premium campaigns.
Marlow clears every unlock requirement easily. All five segments built in an afternoon, and the CSV export set the thresholds: the top 5% starts at £140 of Predicted CLV, and summing that segment sized the VIP programme at £210k of expected next-12-month revenue. Average churn risk exported at 0.58, which is not a crisis, just the one-time-buyer share talking, and it is exactly why the first play below gets built first.
7. The Klaviyo LTV playbook: 5 plays
Everything above becomes revenue here. Five plays, in build order, each using predictive properties as the trigger or the split rather than as decoration.
1. Second-purchase acceleration. Triggered on Placed Order filtered to first-time buyers. A post-purchase branch: education and usage content first, a cross-sell recommendation at roughly 50-70% of your median time-to-second-order, incentive held back as a last-resort final message. KPI: percentage of cohort at 2+ orders by day 90, and median time to second order. Build this first. It attacks the inflection point, and every other play compounds on top of it.
2. EDNO replenishment. A date-property flow on Expected Date of Next Order, starting a few days before the date, or a conditional split in an existing flow where EDNO is within 30 days. One or two messages with a personalised product feed, an optional time-limited incentive on message two only. KPI: conversion rate and share of reorders landing on or before EDNO. For consumable brands this is the highest-ROI flow in the account; the model already knows the customer’s rhythm and you are just showing up on time.
3. Churn-risk intervention. A segment-triggered flow when someone enters “at risk, worth saving.” Branch by value: the high-CLV path gets human-feel outreach and service touches with no discount first, the medium path gets value-reminder content then an offer. Never lead with the discount, or you train your best customers to wait for it. KPI: reactivation rate against a holdout, and margin per reactivated customer. Suppress anyone who purchased in the last 30 days, has an EDNO within 14 days (replenishment owns them), is already in winback, or has been unengaged 180+ days.
4. Predicted-CLV VIP programme. Triggered on entering the Predicted VIP segment. Welcome-to-VIP message, early access to launches and restocks, priority support, review and referral asks, explicitly excluded from discount-led campaigns. KPI: VIP revenue share and VIP retention versus core. Using Predicted rather than historic CLV means you treat tomorrow’s whales as VIPs today, including strong first-order customers the historic tiers would ignore for months.
5. Predictive winback. Triggered on entering “overdue” (EDNO in the past, no order since). Two or three messages replacing your static 60/90-day winback, what’s-new content first, escalating to your strongest offer last, timed per customer instead of one fixed delay. KPI: recovery rate against the old fixed-delay winback, run head to head for a quarter.
The orchestration rule ties it together. Plays 2, 3 and 5 chase overlapping customers, so set the hierarchy once: replenishment beats churn-risk beats winback, enforced with segment exclusions in each flow filter. A customer getting a replenishment nudge, a churn save and a winback in the same week reads as one thing, a brand that does not know them.
Marlow’s build order: second-purchase acceleration first (it attacks the 92-day gap and the 0.58 churn average at source), EDNO replenishment second (a 3-month rhythm is exactly what the model predicts best), then churn intervention, then VIP and predictive winback. The hierarchy is enforced in every flow filter: anyone with an EDNO inside 14 days belongs to replenishment, full stop.
8. Operating cadence: the monthly LTV review
LTV work dies without a ritual. One hour, once a month, same agenda.
Update the cohort table (cumulative margin per customer at 30/60/90/180/365 days by acquisition month, ten minutes if the pipeline exists). Compare curves, not averages, putting the newest complete cohort against same-age cohorts from the previous three months and the same month last year. Check the four leading indicators from section 4. Review the five plays against their KPIs, with holdouts where you have them. Then pick one lever for the month. One. The output of the review is a single decision, not a dashboard tour.
The founder needs three numbers monthly, everything else on request: 60-day margin LTV against new-customer CAC on the latest complete cohort (the “can we afford our growth rate” number), 12-month cohort margin LTV trend (is each new cohort worth more or less than the last), and second-purchase rate by day 90 (is the retention programme compounding).
Four red flags to watch: flattening curves in newer cohorts while old cohorts keep the blended average pretty (the exact failure averages are designed to hide); time to second order stretching three months running; repeat revenue holding while repeat margin falls (discount-dependent retention); and LTV:CAC “improving” because spend was cut rather than because LTV rose, so always check customer count alongside the ratio.
Marlow’s founder report this month: 60-day margin LTV:CAC at 1:1 (stable, spend can step up), 12-month cohort margin LTV trending £64 to £61 (flagged), and 31% of the latest cohort at 2+ orders by day 90 (up two points since the second-purchase flow shipped). The one decision: the dip traces to discount-heavy second orders, so next month’s lever is margin, not frequency.
Want a deeper audit?
This guide covers LTV. An UndergroundEcom audit covers the machine that produces it: full customer journey mapping across email, SMS and automation, campaign and flow performance, and where your retention programme is leaking margin.
30 minutes, operator to operator, no pitch deck.
Or if you want to talk through strategy first: Book a strategy call
Jamie Watkins, CEO at UndergroundEcom
[LinkedIn] | [undergroundecom.com]
